Historical Rainfall Moment Matching in Index Contract Design
Observational StudyClinical Trial
Applying uniform insurance triggers across geography creates spatial adverse selection and pricing distortions due to underlying climate differences. Adjusting village-level trigger dates based on long-run historical rainfall moments maintains equal ex-ante payout probabilities and actuarial unit pricing across heterogeneous locations.
Picture this
Imagine selling umbrella insurance across two towns where one town receives twice as much annual rainfall as the other. If both towns get paid after five inches of rain, everyone in the wet town buys a policy while the dry town ignores it. By setting the payout trigger at ten inches for the wet town and five inches for the dry town, the insurer keeps the odds of receiving a payout identical for both towns, allowing the policy to be sold at the exact same price everywhere.
What the evidence says
Historical mean daily rainfall (4.18 mm vs 4.12 mm, t = -0.11) and historical coefficient of variation (0.868 vs 0.845, t = -0.16) were statistically indistinguishable between payout and non-payout villages, confirming that 2011 payouts were driven by exogenous weather shocks rather than baseline risk differences.
- Who was studied
- Historical rainfall data (1999–2006) across 63 sample villages in Andhra Pradesh, Uttar Pradesh, and Tamil Nadu, India.
- How
- T-tests comparing historical rainfall moments (mean and coefficient of variation) across 4 payout villages and 38 non-payout villages.
What to do
Tailor weather index trigger thresholds to local historical rainfall distributions to maintain uniform actuarial pricing and prevent spatial selection bias across implementation sites.
From the source
"AICI tailored the insurance contract design details for each village on the basis of that village's historical rainfall distribution... to keep the unit price of insurance as similar as possible across villages, and the trigger dates were therefore adjusted to keep payout probabilities constant."
Risk, Insurance and Wages in General Equilibrium